Part of: this research — view all 12 rows
Current Status and the Observed-Progress Hurdle As of November 14, 2026, no top-five developer has issued a qualifying statement. Anthropic's November 10 SEC filing is the closest to date, declaring a determination that it met its Responsible Scaling Policy's AI R&D threshold. Crucially, however, the company explicitly based this on "what we expect from future models and our overall rate of progress, rather than on what we have observed from current models" Anthropic, PBC - S-1/A-2 Anthropic S-1/A: AI Research Threshold Met, IPO at $70-$78 Anthropic says it met its AI research threshold. A determination resting on expectations does not confirm an observed doubling of the overall rate of research progress, leaving the resolution criteria unmet. Other metrics, such as >90% of codebase contributions or agent-hours surpassing human-hours, measure coding throughput rather than the required overall research uplift in value.
Anthropic's Upward Trajectory Anthropic remains the most likely first mover, with its measured capability growth rapidly approaching the threshold. The company's own capability-index fits have risen from 1.53× in September to 1.88× in the Fable 5.2 and Mythos 5.2 system card Claude Fable 5.2 & Claude Mythos 5.2 System Card. Furthermore, METR's independent assessment found capability progress running at roughly 1.9–2× the baseline rate, attributing about 80% of that acceleration to AI Independent Assessment of AI R&D Acceleration .... While an independent evaluator's finding cannot substitute for a developer statement, it confirms the numerical proximity. Anthropic has promised to publish a dedicated report on overall trends in acceleration "in the coming weeks" Claude Fable 5.2 & Claude Mythos 5.2 System Card AI R&D Is Accelerating Towards Potential RSI in 2027 Roadshow Monday: The Discount, the Unreleased Model, and the Trigger ..., making late 2026 and early Q1 2027 highly sensitive periods for a potential announcement. These reporting opportunities support a 10% probability of a qualifying statement by January 1, 2027, and 25% by February 20, 2027, but numerical proximity alone does not make an early statement the central case.
The Rest of the Frontier Other developers currently trail on this specific threshold. Google DeepMind's recent Gemini 4 Argon report found no significant end-to-end acceleration of AI R&D, though Sundar Pichai noted an expectation of continued acceleration from recursive self-improvement over the next year. OpenAI rates all its current models below "High" for AI self-improvement and has not published a multiplier for overall research progress, despite reporting internal milestones and a highly capable unreleased model. Consequently, near-term probability rests overwhelmingly on Anthropic.
Governance, Disclosure, and the Long Tail Despite Anthropic's numerical proximity, a clean public statement that progress has "at least doubled" is not guaranteed in the near term. As a newly listed company, Anthropic faces heightened legal and disclosure scrutiny, which may incentivize careful hedging—such as continuing to cite "expected" progress or stating that no single model clears the bar individually. Translating capability-index slopes into realized organizational value uplift also remains ambiguous Independent Assessment of AI R&D Acceleration ... Task Substitution and Uplift; if a 1.9× total acceleration is only 80% AI-attributable, the true AI uplift is closer to 1.5×, explaining the company's persistent caution. Broader structural bottlenecks in translating AI capability gains into measurable R&D productivity, together with governance friction delaying explicit public attribution, make a lag between approaching the capability threshold and issuing a qualifying statement more likely. I therefore place the median at June 1, 2027, rather than treating the imminent report as the most likely resolution point. Finally, deliberate pacing, safety-driven pauses, or new inference-time safeguards could intentionally suppress the observed rate below 2×. These measurement, governance, and disclosure friction points support a 75% probability by March 1, 2028, while leaving a long upper tail, with 90% probability reached only by June 30, 2030.
Accounting for broader expectations around structural bottlenecks in translating AI capability gains into measurable R&D productivity and governance friction that could delay explicit public attributions pushed the median timeline slightly later into mid-2027.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited